What denoise strength should I start with?

Run 0.3, 0.5 and 0.7 with the same prompt before changing any words. The right value is usually obvious immediately and it saves rewriting a prompt that was never the problem.

Why did the output ignore my source image?

Denoise strength was too high. Above roughly 0.7 the source is buried under noise and the model reconstructs mostly from your prompt.

Why did the output ignore my prompt?

Strength was too low. Below about 0.25 there is not enough noise for the model to meaningfully change anything.

Can I keep a face and still restyle heavily?

Not with denoise strength alone — the two requirements conflict. Identity preservation across a large change needs reference conditioning instead.

Is denoise strength the same as guidance scale?

No. Denoise strength controls how much source survives; guidance scale controls how literally the model follows the prompt. They are independent.

Does it apply to text-to-image?

No. With no source image there is nothing to preserve, so the concept only exists in image-to-image and inpainting.

AI image glossary

What is denoise strength?

In image-to-image, how much of your source image is discarded before regeneration. Low values preserve the original; high values treat it as a loose reference.

Denoise strength is the single most important control in image-to-image generation. It sets how much noise is added to your source before the model denoises it back toward your prompt — which decides how much of the original survives. At 0.2 the photograph is essentially intact; at 0.85 it is little more than a mood board.

Five paper cards in a row, progressively obscured by grey speckle from left to right

The mechanism explains the behaviour

Image-to-image adds noise to your source and then denoises toward the prompt. Add a little and the original structure remains visible underneath, so the model largely preserves it. Add a lot and the source is buried, so the model reconstructs mostly from your words.

The three ranges worth knowing

Below about 0.35 surfaces and colour change while composition and identity hold. Between 0.4 and 0.6 composition survives but content is renegotiated, which is where most useful restyling happens. Above 0.7 the source is a suggestion rather than a constraint.

Most bad results are the wrong number, not the wrong words

If the output ignored your source, strength was too high. If it ignored your prompt, too low. Sweeping three or four values with one unchanged prompt tells you more in a minute than an hour of rewriting, and it is the first thing to try when a result disappoints.

What this control cannot do

Preserving a specific face requires low strength, but meaningfully restyling requires high strength, and those two requirements are in direct conflict. That conflict is precisely why identity preservation needs reference conditioning rather than a strength setting.

Denoise strength at a glance

Range: 0 to 1
Near 0: Almost the source image
Near 1: Almost ignores the source
Applies to: Image-to-image only
Common start: 0.5 to 0.7
Also called: Strength, image weight

Related terms

What denoise strength sits next to

inpainting explainedChanging one region only.

What is outpainting?Adding beyond the frame.

start from a photoWhere this control lives.

AI photo editingEditing, prompt-driven.

seeds explainedFixing the starting noise.

AI Image GlossaryEvery term, defined.

Put it to use

Where this control appears

generate from a descriptionThe main generator — text to finished image.Open browse real prompts120 images with the exact prompts that made them.Open Image to ImageStart from a picture you already have.Open AI Photo EditorChange part of a photo, keep the rest.Open AI Image UpscalerEnlarge to 4K without the smearing.Open enhance a photoExposure, noise and old-photo restoration.Open masked object removalMasked removal that rebuilds the background.Open blur a photo backgroundDepth-aware, not a uniform filter.Open Colorize PhotoPlausible colour for black and white.Open generate a logoMarks that still read at 16 pixels.Open design a tattooDesigns that survive being tattooed.Open painting and illustrationIllustration and painting, not photographs.Open AI Headshot GeneratorStudio-grade portraits without a studio.Open the photo specificationThe exact specification, and how to meet it.Open AI Interior DesignRestyle a room you can photograph.Open terms explainedSeed, denoise, inpainting, explained.Open

Denoise strength — questions people ask

What denoise strength should I start with?: Run 0.3, 0.5 and 0.7 with the same prompt before changing any words. The right value is usually obvious immediately and it saves rewriting a prompt that was never the problem.
Why did the output ignore my source image?: Denoise strength was too high. Above roughly 0.7 the source is buried under noise and the model reconstructs mostly from your prompt.
Why did the output ignore my prompt?: Strength was too low. Below about 0.25 there is not enough noise for the model to meaningfully change anything.
Can I keep a face and still restyle heavily?: Not with denoise strength alone — the two requirements conflict. Identity preservation across a large change needs reference conditioning instead.
Is denoise strength the same as guidance scale?: No. Denoise strength controls how much source survives; guidance scale controls how literally the model follows the prompt. They are independent.
Does it apply to text-to-image?: No. With no source image there is nothing to preserve, so the concept only exists in image-to-image and inpainting.

Related AI image terms

Start creating today

Generate your first image with the best AI image generator.

Turn a sentence into a finished, photorealistic image in seconds — then refine every detail. No setup, no Discord, no GPU.

Try denoise strength yourself Get Started

Join 4,200+ creators using LaFoto